Global sensitivity analysis of elastomer joints between substructures using random balance design method

摘要

Plug-In Digital Framework is a system response analysis tool when the system components are composed of black-box modules. Generally, the dynamic characteristics of joints between the system components affect largely on system responses, and have displacement- and frequency-dependent stiffness and loss factor. Thus, the sensitivity of each joint parameters should be estimated in global sense. In this study, we introduce a global sensitivity analysis procedure under the Plug-In Digital Framework. To efficiently calculate the system responses, the FRF-based substructuring method is introduced. Global first-order sensitivity is obtained from the analysis of variance for system responses calculated at randomly sampled points. The random sample points are generated by the random balance design (RBD) which uses equal space sampling in cumulative distribution function for a parameter and random permutation. The Fourier transform for the sorted responses for a parameter provides the global first-order sensitivities of each joint stiffness and loss factor. The proposed global sensitivity analysis method are applied to an interior noise problem of a passenger car and the efficiency of the global sensitivity analysis method is discussed.